© 2018. Published by The Company of Biologists Ltd | Biology Open (2018) 7, bio035576. doi:10.1242/bio.035576

RESEARCH ARTICLE Divergent Hemogen of teleosts and mammals share conserved roles in erythropoiesis: analysis using transgenic and mutant zebrafish Michael J. Peters, Sandra K. Parker, Jeffrey Grim*, Corey A. H. Allard‡, Jonah Levin§ and H. William Detrich, III¶

ABSTRACT (Krüger et al., 2002, 2005). Here we analyze the developmental Hemogen is a vertebrate transcription factor that performs important roles of teleost Hemogen using the zebrafish model system and its functions in erythropoiesis and testicular development and may powerful suite of reverse-genetic technologies. contribute to neoplasia. Here we identify zebrafish Hemogen and Teleost Hemogen was discovered using a subtractive show that it is considerably smaller (∼22 kDa) than its human ortholog hybridization screen designed to isolate novel erythropoietic (∼55 kDa), a striking difference that is explained by an underlying genes from fish belonging to the largely Antarctic suborder modular structure. We demonstrate that Hemogens are largely Notothenioidei (Detrich and Yergeau, 2004; Yergeau et al., composed of 21-25 amino acid repeats, some of which may function 2005). Sixteen species belonging to the icefish family as transactivation domains (TADs). Hemogen expression in embryonic (Channichthyidae) are unique among vertebrates because they are and adult zebrafish is detected in hematopoietic, renal, neural and white-blooded; they fail to execute the erythroid genetic program or gonadal tissues. Using Tol2- and CRISPR/Cas9-generated transgenic produce hemoglobin (Cocca et al., 1995; Near et al., 2006; Zhao zebrafish, we show that Hemogen expression is controlled by two et al., 1998). Forty-five candidate erythropoietic cDNAs were Gata1-dependent regulatory sequences that act alone and together to recovered using representational difference analysis (Hubank and control spatial and temporal expression during development. Partial Schatz, 1999) applied to kidney marrow transcriptomes of two depletion of Hemogen in embryos by morpholino knockdown reduces notothenioid species, one red-blooded and the other white-blooded the number of erythrocytes in circulation. CRISPR/Cas9-generated (Detrich and Yergeau, 2004; Yergeau et al., 2005). One of the zebrafish lines containing either a frameshift mutation or an in-frame unknown genes, clone Rda130, was similar to mammalian deletion in a putative, C-terminal TAD display anemia and embryonic Hemogen and was expressed only by the red-blooded notothenioid. tail defects. This work expands our understanding of Hemogen and Although Hemogen is clearly involved in hematopoiesis, its provides mutant zebrafish lines for future study of the mechanism of mechanism remains incompletely understood. In human cell lines, this important transcription factor. Hemogen activates erythroid transcription in part by recruiting the histone acetyltransferase P300 to acetylate Gata1 (Zheng et al., KEY WORDS: Anemia, CRISPR/Cas9, Gene editing, Hematopoiesis, 2014). Like Gata1, Hemogen protects erythroid cells from apoptosis Transcription by upregulating anti-apoptotic factors (e.g. Nf-κB, Bcl-xL) that are critical for terminal differentiation (Li et al., 2004; Rhodes et al., INTRODUCTION 2005; Zhang et al., 2012). Hemogen (Hemgn) is a vertebrate transcription factor that is The regulation of Hemogen expression is of interest because it is expressed in mammalian hematopoietic progenitors (Lu et al., 2001; overexpressed frequently in patients with a variety of cancers and Yang et al., 2001) and has been implicated in erythroid leukemias (An et al., 2005; Forbes et al., 2017; Li et al., 2004). This differentiation and survival (Li et al., 2004). Originally identified putative oncogene, which is located in a human chromosomal in mice and subsequently described in humans as EDAG region (9q22) of leukemia-associated breakpoints, has been linked (Erythrocyte Differentiation Associated Gene), Hemogen has also to proliferation and survival of leukemic cells and to induction of been implicated in testis development in mammals and chickens tumor formation in mice (Chen et al., 2016; Lu et al., 2002). Thus, (Nakata et al., 2013; Yang et al., 2003), and in osteogenesis in rats somatic mutations in Hemogen or its regulators may contribute to neoplasia. The zebrafish is a well-established model organism for studying Department of Marine and Environmental Sciences, Northeastern University, Nahant, MA 01908, USA. hematopoiesis in vertebrates because it produces the same blood *Present address: Department of Biology, The University of Tampa, Tampa, FL lineages as mammals (de Jong and Zon, 2005; Paffett-Lugassy and ‡ 33606, USA. Present address: Department of Biochemistry and Cell Biology, Zon, 2005). In zebrafish, erythropoiesis occurs in sequential waves at Geisel School of Medicine at Dartmouth College, Hanover, NH 03755, USA. §Present address: Department of Biochemistry, McGill University, Montreal, unique anatomical locations in embryos and adults that correspond to Quebec H3G1Y6, Canada. analogous sites in mammals (Galloway and Zon, 2003). Many of the

¶Author for correspondence ([email protected]) molecular players that orchestrate the erythroid program appear to be conserved between zebrafish and mammals, but relatively few have M.J.P., 0000-0003-4932-528X; J.G., 0000-0002-3344-0657; C.A.H.A., 0000- been functionally characterized in zebrafish. Nevertheless, mutant 0002-3554-2059; H.W.D., 0000-0002-0783-4505 zebrafish models accurately phenocopy human blood diseases caused This is an Open Access article distributed under the terms of the Creative Commons Attribution by mutations in major erythroid factors, such as Gata1 (Lyons et al., License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution and reproduction in any medium provided that the original work is properly attributed. 2002) and Erythroid beta-spectrin (Liao et al., 2000). The purpose of this study is to characterize the regulation of

Received 7 May 2018; Accepted 16 July 2018 Hemogen expression and the function of the Hemogen in Biology Open

1 RESEARCH ARTICLE Biology Open (2018) 7, bio035576. doi:10.1242/bio.035576

Fig. 1. Zebrafish Si:dkey-25o16.2 and human Hemogen are orthologous and encode related that differ in size. (A) Structure of the zebrafish Hemogen-like gene, Si:dkey-25o16.2. Two conserved noncoding elements (C1 and C2, black boxes) were identified in a 2 kb segment proximal to the start codon (see Results, Figs 4-6). Coding exons, white boxes; noncoding exons, gray boxes. Numbers indicate length in bp. (B) Synteny of loci for zebrafish Si:dkey-25o16.2 on 1 and Hemogen on human (region q22). Transcriptional orientations indicated by arrows. (C) Alternative splicing of zebrafish Hemogen-like transcripts showing sequenced regions. Introns are shown as chevrons. Transcripts 1 and 2 differ by retention of 12 bp of intron (red). (D) Modular structures of zebrafish and human Hemogen proteins each encoded by four exons (numbered boxes). Locations of truncating mutations found in some human cancers (Forbes et al., 2017) are indicated by asterisks. Predicted regions and motifs: green, coiled coil; blue, nuclear localization signal; red, four residues introduced by alternative splicing; yellow, tandem peptide repeats; brown, acidic repeat with transactivation domain (TAD) motif; gray, no prediction. (E) Three-dimensional ab initio models of Hemogens. The ribbon diagram of the zebrafish protein, color-coded as in panel D, is superimposed on the gray, space-filling model for the human protein (See Materials and Methods). zebrafish. We identify the zebrafish Hemogen ortholog, which predicted gene Si:dkey-25o16.2 on chromosome 1 of the zebrafish despite being only 40% as large as the human protein, contains genome (Howe et al., 2013). When we compared the synteny of the similarly arranged functional motifs. Hemogen is expressed in putative teleost and mammalian orthologs, represented in Fig. 1B blood, testis, ovaries, kidneys and the central nervous system in by zebrafish Si:dkey-25o16.2 (chromosome Dr1) and human zebrafish. Two tissue-specific, alternative Hemogen promoters Hemogen (chromosome Hs9), we found that the flanking genes are associated with conserved noncoding elements (CNEs) and and their transcriptional orientations were conserved, which have distinct regulatory functions in primitive and definitive strongly supported Si:dkey-25o16.2 as the zebrafish Hemogen hematopoiesis and other processes. By analysis of morphant and ortholog. mutant zebrafish, we show that Hemogen is required for normal erythropoiesis and that this role depends in part on a cluster of acidic Structure of the zebrafish Hemogen gene residues within a putative, C-terminal transactivation domain The basic structure of the Hemogen gene of teleosts and mammals (TAD). was also found to be highly conserved; four coding exons were separated by three introns (Fig. 1A), and two introns were found in RESULTS the 5′-UTR. Two transcription start sites were predicted to occur Teleosts contain a single Hemogen-like gene that is syntenic within 2 kb upstream of the Hemogen start codon in zebrafish with human Hemogen (Fig. 1A) and these appear to correspond to the hematopoietic- and Chromosomal synteny is an important criterion when assigning testis-specific Hemogen promoters (noncoding exons 1H and 1T, gene relationships across divergent taxa. Despite the whole- respectively) described for mammals (Yang et al., 2003). Alignment genome duplication (WGD) that coincided with the separation of Hemogen genes from ten teleost species (Yates et al., 2016) of teleosts from more basal ray-finned fishes and tetrapods revealed two conserved non-coding elements, CNE1 and CNE2, (Postlethwait et al., 2000), the sequenced genomes of nearly all that overlapped with zebrafish exons 1T and 1H, respectively fishes retain a single Hemogen-like gene. We cloned zebrafish (Fig. 1A). We hypothesize that these elements function individually

Hemogen-like cDNAs and found that they corresponded to the or together to regulate transcription of Hemogen. Biology Open

2 RESEARCH ARTICLE Biology Open (2018) 7, bio035576. doi:10.1242/bio.035576

Transcription of the zebrafish Hemogen gene yields multiple 2010). When the two models were superimposed, amino acid mRNA isoforms sequences shared by human and zebrafish Hemogens showed 98% We confirmed transcription from both promoters in zebrafish by coincidence and a TM-score of 0.71. The N-termini of the zebrafish isolating and sequencing four splicing variants (Fig. 1C). Three and human Hemogens presented exposed CC domains that isoforms were transcribed from the proximal promoter (exon 1H, may serve as binding sites for Gata1 (Zheng et al., 2014). The Fig. 1A,C), each containing the same 5′-untranslated region ‘disordered’ C-termini of Hemogens from zebrafish and humans (5′-UTR), compared to two corresponding mammalian transcripts. were comprised of two distinct elements: proline-rich repeats Alternative splicing of the second coding exon produced transcripts 1 (yellow) and an acidic, C-terminal repeat containing the TAD motif and 2, which differ by four additional codons in the latter (Fig. 1C, (brown) (Fig. 1E; Fig. S1). The former may coalesce as rigid linkers red); the shorter version has not been described in mammals. to extend the TAD motif to binding partners. These features are Transcript 3 retained the entire third intron (156 bp), which common to transcription factors, as epitomized by the structure of introduced a premature translation-termination codon. A fourth p53 (Wells et al., 2008). isoform was transcribed from the distal promoter (1T) located ∼1.65 kb upstream of the translation start codon (Fig. 1A,C). Splicing Hemogen expression tracks the ontogenetic progression of of exons 1T and 1H to form the 5′-UTR of transcript 4 made use of hematopoiesis in zebrafish canonical donor (AT-GT) and acceptor (AG-TT) splice sites. The spatial and temporal patterns of Hemogen expression were evaluated in zebrafish between 2 and 144 h post fertilization (hpf) Teleost and mammalian Hemogen proteins differ markedly in by whole-mount in situ hybridization (WISH) (Fig. 2A-H). size but share structural motifs Hemogen transcripts were not apparent prior to somitogenesis Teleost Hemogen-like genes encoded shorter proteins (194-289 (Fig. 2A) but first appeared at the ten-somite stage in punctate, amino acids) than the annotated Hemogen genes of mammals (417- intersomitic foci in the lateral plate mesoderm (LPM; Fig. 2B). By 827 amino acids), and the overall amino acid sequence similarity 20 hpf, Hemogen was expressed throughout the intermediate cell between teleost and mammalian orthologs was modest (18%-38%). mass (ICM) and posterior blood island (PBI) (Fig. 2C), the sites of Despite this heterogeneity in length and sequence, Hemogens of primitive hematopoiesis (Bertrand et al., 2007; Davidson and Zon, teleost fish and mammals shared predicted structural motifs, as 2004). Primitive erythrocytes expressed Hemogen as they entered shown in Fig. 1D,E for zebrafish (198 aa, 22 kDa) and human (484 circulation at 33 hpf (Fig. 2D). aa, 55 kDa) orthologs, respectively. Their N-termini (zebrafish Definitive hematopoiesis in zebrafish embryos commences in residues 1-74, human 1-78) were substantially conserved (51% the aorta gonad mesonephros (AGM) region at 30 hpf with the sequence similarity; Fig. S1) and contained two predicted coiled- emergence of hematopoietic stem progenitor cells (HSPCs) that coil (CC) forming alpha-helices, the second of which was a putative subsequently seed the caudal hematopoietic tissue (CHT) and the nuclear localization signal (NLS) (Yang et al., 2001) (Fig. 1D; thymus (Murayama et al., 2006). By 144 hpf, HSPCs migrate from Fig. S1). By contrast, their C-termini (zebrafish residues 75-198, the CHT to establish a niche associated with the pronephric glomeruli human 79-484) were weakly conserved in sequence (13% (Bertrand et al., 2008). Although we did not detect Hemogen mRNA similarity), but both were rich in Pro and Glu residues (Figs S1 in the AGM (Fig. 2D), we observed strong expression in cells of the and S2), consistent with intrinsic disorder of these regions (Dyson CHT at 48 and 144 hpf (Fig. 2E,G) and in the region of the pronephric and Wright, 2005). Furthermore, the C-termini shared modular glomeruli at 144 hpf (Fig. 2E,F). In the adult zebrafish kidney, structures – each was built of several 21-25 amino acid motifs, three Hemogen was strongly expressed in progenitor cells in the interstitial in zebrafish and nine in humans, with distinct but related hematopoietic stem cell niche between pronephric tubules (Fig. 2H). consensus sequences (PEXXXIAEXXXXXQEVXPQXXLVP and Hemogen expression was robust in progenitors but absent in mature YSXEXYQEXAEPEDXSPETYQEIPX, respectively) (Fig. 1D,E; erythrocytes (Fig. 2H), whereas an anti-sense riboprobe for βe1- Figs S1 and S2). Thus, the size heterogeneity between zebrafish globin hybridized exclusively to mature erythrocytes but not to and human Hemogens was largely attributable to the number of progenitor cells (data not shown). repetitive segments contained within each. Hemogen has been shown to function as a nuclear transcription Within the C-termini of teleost Hemogens, we identified a factor in mammals (Zheng et al., 2014). To determine whether or conserved acidic region (zebrafish residues 119-169, 35-49% not Hemogen is likely to play the same role in zebrafish, we similarity across ten species) that was similar to an acidic region examined Tg(Lcr:EGFP)cz3325Tg embryos at 48 hpf by indirect of the mouse protein (Yang et al., 2001). Given the transactivation immunofluorescence microscopy using an antibody specific for functions of Hemogen in humans (Zheng et al., 2014), we Hemogen. Tg(Lcr:EGFP)cz3325Tg zebrafish have been used to track investigated whether the zebrafish and human proteins possessed both primitive and definitive erythrocytes (Ganis et al., 2012). TAD motifs based on the consensus sequences φφxxφ or φxxφφ, Fig. 2I shows that Hemogen accumulated in the nuclei (red signal) where φ is a bulky hydrophobic residue (Dyson and Wright, 2016). of GFP-labeled circulating erythrocytes in the dorsal aorta, thus, its The acidic C-termini of both Hemogens contained one TAD motif. role in transcription is likely to be conserved in zebrafish. Four additional TAD motifs were distributed in other regions of the human protein (Fig. S1). Alternative promoters regulate Hemogen expression in To assess the three-dimensional conformations of zebrafish and zebrafish hematopoietic and reproductive tissues human Hemogens, although in a static context, we generated ab In zebrafish, we also detected Hemogen expression in the initio tertiary structural models with I-Tasser (Yang et al., 2015) hindbrain and in the pronephric tubules of embryonic zebrafish using the best of ten predicted templates (Fig. 1E, see Materials and between 30 and 48 hpf (Fig. 3A,B) and in adult zebrafish brain Methods). The structures for zebrafish and human Hemogens and reproductive tissues (Fig. 3C-H). The alternative Hemogen had template modeling scores (TM-scores) of 0.45 and 0.55, promoters found in zebrafish probably correspond to the respectively, where a TM-score >0.3 indicates significantly hematopoietic and testis-specific Hemogen promoters of different (P<0.001) from random structures (Xu and Zhang, mammals (Yang et al., 2003). To quantify relative levels of Biology Open

3 RESEARCH ARTICLE Biology Open (2018) 7, bio035576. doi:10.1242/bio.035576

Fig. 2. Hemogen expression in zebrafish embryos. (A-H) Wild-type embryos, WISH. (A) Epiboly at 9 hpf. Hemogen expression was not detected. (B) Ten-somite stage. Hemogen transcripts along the lateral plate mesoderm (LPM). (C) 20 hpf. Hemogen staining in the intermediate cells mass (ICM) and posterior blood island (PBI). The inset shows a sense probe control. (D) 33 hpf. Hemogen-positive primitive erythrocytes of the peripheral blood (PB) exited the Ducts of Cuvier (DC) onto the yolk. Staining at the midbrain-hindbrain boundary (MHB) was observed. (E) 144 hpf. Hemogen expression in the caudal hematopoietic tissue (CHT) and pronephric kidney (PK) and in erythrocytes in the heart (H). The asterisk indicates the plane of the cross-section in panel F. (F) 144 hpf. Cross-section of embryo in panel E showing heavily stained pronephric ducts. (G) 48 hpf. Lateral aspect of tail. Hemogen transcripts in the CHT and pronephric tubule duct (PD). (H) Kidney touch print from adult fish. Hemogen expression was observed in proerythroblasts (ProE) and normoblasts (N) but not in erythrocytes (E). (I) 48 hpf. View of circulating EGFP+ erythrocytes in the dorsal aorta (DA) of Tg(Lcr:EGFP)cz3325Tg zebrafish after staining for Hemogen protein by indirect immunofluorescence. Hemogen (red signal) accumulated in nuclei (Nu) of erythrocytes whereas the cytoplasm (C) was marked by EGFP. Other abbreviations: AGM, aorta gonad mesonephros; PT, pronephric tubule; CV, caudal vein; DA, dorsal aorta; G, gut; M, myotomes; NC, notochord; SB, swim bladder; SC, spinal cord. Scale bars: (A-F) 250 µm; (G) 1 mm; (H) 100 µm; (I) 50 µm.

from exon 1T, one must infer transcription from the proximal promoter by difference. Transcription from the proximal promoter was greatest in peripheral blood; the presence of transcripts from this promoter in testis and ovarian tissue may be due to contaminating blood RNA. The distal promoter was highly active in both peripheral blood and in testes but not in ovaries.

Hemogen CNEs are predicted targets for transcription factors that regulate erythropoiesis and spermatogenesis In teleosts, we identified two evolutionarily conserved non-coding elements, CNE1 and CNE2, that were tightly associated with exons 1T and 1H, respectively (Figs 1A and 4A). These elements may function as core promoters and/or enhancers to regulate transcription of the different Hemogen isoforms in zebrafish. To identify potential regulators of Hemogen transcription, we used ConTra v2 (Broos et al., 2011) to predict transcription factor binding motifs in the aligned Hemogen CNEs from two mammals and nine teleosts (Yates et al., 2016) (Fig. 4B,C). Each CNE contained binding motifs for transcription factors involved in erythropoiesis and/or spermatogenesis. In zebrafish CNE2, two Gata1 binding sites, located +59 and +127 bp downstream relative to the transcription start site, aligned with Gata1 sites known to be active in the mammalian Hemogen promoter (Fig. 4C) (Yang et al., 2006). Each Gata motif was paired with a predicted E-box; this motif in Hemogen CNE2 is a known target of the Ldb1-erythroid-complex recruited by Scl (Soler et al., 2010). CNE2 also contained binding sites for Klf4, a driver of zebrafish primitive erythropoiesis (Gardiner et al., 2007), for Myb, a regulator of zebrafish definitive hematopoiesis (Soza-Ried et al., 2010), and for HoxB4, a regulator of Hemogen expression in mammalian hematopoietic stem cells (Jiang et al., 2010). The distal CNE1 of teleosts possessed a similar suite of transcription factor binding motifs in roughly the same arrangement as the proximal CNE but with the notable addition of binding sites for Sox9 and the Androgen receptor (Fig. 4B), both of which play roles in zebrafish spermatogenesis (Hossain et al., 2008; Rodríguez-Marí et al., 2005). CNE1, like CNE2, contained transcription from each promoter in zebrafish (Fig. 3I), we pairs of E-box and Gata motifs downstream of the zebrafish performed qRT-PCR on total RNA from adult peripheral blood, transcription start site (+15 and +48 bp, respectively). CNE1 may testis and ovaries (Fig. 3J) using primer pairs specific for exons 1H function as an enhancer for the Hemogen gene and/or act as the core and 1T. Because all of exon 1H was included in transcripts initiated promoter for exon 1T. Biology Open

4 RESEARCH ARTICLE Biology Open (2018) 7, bio035576. doi:10.1242/bio.035576

Fig. 3. Alternative promoters drive Hemogen expression in hematopoietic and nonhematopoietic tissues in zebrafish. WISH of wild- type embryos (A-B) and adult tissues (C-H). (A) 48 hpf. Hemogen expression in the pronephric kidney glomeruli (PG), pronephric tubule duct (PD), caudal hematopoietic tissue (CHT) and brain (Br). (B) 48 hpf. Section showing strong Hemogen expression in the hindbrain (HB) but at low levels in the midbrain (MB). (C,D) Dorsal (C) and ventral (D) views of the adult zebrafish brain after staining for Hemogen transcripts. CC, crista cerebellaris; Hy, hypothalamus; EG, eminentia granularis. (E) Schematic drawing of the dorsal view. Hemogen was highly expressed at the midbrain- hindbrain boundary within the EG, in the CC and in the Hy. The asterisk indicates the plane of the cross-section in panel F. (F) Section of the hindbrain showing Hemogen expression in the periventricular gray zone (PGZ). (G) Hemogen was expressed by Sertoli cells (SE) between the seminiferous tubules (ST) of the testes. (H) Hemogen was expressed in early (I-III) but not late (IV) stage oocytes. Transcripts accumulated around the germinal vesicle (GV). (I) Schematic of the Hemogen noncoding exons 1T and 1H (gray) upstream of the first coding exon (white); bent arrows, transcription initiation sites. Arrowheads mark primer binding sites for qPCR amplification of transcripts initiated from exons 1T or 1H. (J) Expression of transcripts from alternative promoters determined by qRT-PCR using RNA from blood, testes and ovaries of adult TU zebrafish. Expression in three biological replicates were normalized to β-actin and calculated relative to ovaries. Error bars represent the standard deviation. Transcription initiated from 1H must be inferred by difference [1H – 1T] because the 1H primers also amplified 1T transcripts. Other abbreviations: Ce, corpus cerebelli; MO, medulla oblongata; OB, olfactory bulb; OT, optic tectum; SR, superior raphe; Te, telencephalon; TS, torus semicircularis. Scale bars: 250 µm (A,B,F-H); 1 mm (C,D).

CNE2, are important regulators of Hemogen expression in zebrafish erythrocytes. We performed WISH to compare the expression of Hemogen and Embryonic beta-globin (βe1-globin) in embryos produced by the Gata1-null mutant, vlad tepes (vltm651) (Lyons et al., 2002). At 33 hpf, Hemogen was expressed normally in circulating blood cells and in the hindbrain of wild-type siblings (Fig. 5B), and βe1-globin was abundant in the blood (Fig. 5B, inset). Homozygous vltm651 mutant siblings, by contrast, failed to express Hemogen in the blood and brain (Fig. 5C). This result mimicked the loss of βe1-globin in vltm651 mutants, with the exception that βe1-globin expression persisted in the PBI (Fig. 5C, inset), as has been demonstrated for α1-globin, Scl and Gata1 (Jin et al., 2009).

Tg(Hemgn:mCherry) zebrafish reveal the functions of the two Hemogen promoters To determine the tissue-specific regulatory profiles of the two Hemogen promoters, we generated transgenic zebrafish embryos [Tg(Hemgn:mCherry,myl7:EGFP)] in which the mCherry reporter was controlled by the putative promoter elements (Fig. 6). The dual promoter, P1 (2248 bp), spanned the upstream, non-coding region to Hematopoietic and neural expression of Hemogen in the Hemogen start codon and contained both CNEs. Transgenic fish zebrafish is dependent on Gata1 binding to the promoter were outcrossed to wild-type TU zebrafish and offspring with the CNEs strongest mCherry expression were selected as founders. In the early In mammals, transcription of Hemogen from the proximal embryo, the P1 transgene drove expression of mCherry in primitive promoter is tightly regulated by Gata1 in hematopoietic cells blood cells of the ICM and the PBI (20 hpf, Fig. 6B) and in primitive (Yang et al., 2006). To investigate whether Gata1 regulates erythrocytes in circulation (Movie 1). Between 2 and 8 dpf, mCherry Hemogen in zebrafish, we analyzed a Gata1 ChIP-seq dataset that was expressed strongly throughout the pronephric ducts (Fig. 6C) and was generated to assess Gata1 activity in adult zebrafish was present in the proximal convoluted tubule at 72 hpf (Fig. 6D). In erythrocytes (Yang et al., 2016). Fig. 5A shows that Gata1 adult transgenic fish, the head and trunk kidneys were positive for the bound to CNE1 and CNE2 at sites overlapping their Gata motifs reporter (Fig. 6H), as were Sertoli cells surrounding the seminiferous (red lines), which indicates strongly that Gata1 is required for tubules of the testes (Fig. 6I). Therefore, the ∼2.2 kb P1 transgene transcription of Hemogen in zebrafish. Corroboration that CNE1 contained all of the regulatory elements necessary to recapitulate and CNE2 were active chromatin regions was provided by ATAC- Hemogen expression (Fig. 6B-I). We note that the dual promoter did seq and DNase I hypersensitive site analysis (Yang et al., 2016) not confer detectable ovarian or neural expression, which may require

(Fig. 5A). Our data reveal that Gata motifs in CNE1, like those in more distal sequences. Biology Open

5 RESEARCH ARTICLE Biology Open (2018) 7, bio035576. doi:10.1242/bio.035576

Fig. 4. Conserved elements in the zebrafish Hemogen promoter are predicted targets for transcription factors. (A) Schematic of the zebrafish Hemogen gene. CNEs, black; coding exons, white; noncoding exons, gray; transcription initiation sites, bent arrows. Numbers indicate length in bp. (B,C) Sequence alignments of CNE1 and CNE2, respectively, from nine teleost species, mice and humans. ConTra software (Broos et al., 2011) predicted transcription factor binding sites for the Androgen receptor (light green), Brca1 (cyan), Foxl2 (pink), Gata1 (dark blue), Gfi1 (orange), HoxB4 (sky blue), Hnf1a (dark green), Klf4 (yellow), Myb (dark gray), P300 (red), Sox9 (purple) and Scl/Lmo2/Ldb1 complex (light gray). Splice donor sites are highlighted black. Species abbreviations: Dr, Danio rerio; Cs, Cynoglossus semilaevis; Gm, Gadus morhua; Ga, Gasterosteus aculeatus; Ol, Oryzias latipes; Xm, Xiphophorus maculatus; On, Oreochromis niloticus;Tr,Takifugu rubripes; Tn, Tetraodon nigroviridis; Mm, Mus musculus; Hs, Homo sapiens.

We found that the same expression profile was driven by the supports prior observations that Hemogen expression is limited endogenous Hemogen promoter in embryonic zebrafish by using to primitive erythrocytes and immature definitive progenitors CRISPR/Cas9 technology to insert the mCherry gene (containing a (Lu et al., 2001). polyadenylation motif) two codons downstream of, and in frame with, the Hemogen start codon (See Materials and Methods; Fig. S3A,C). Hemogen promoters have different functions in primitive and Homology-directed integration of the transgene, confirmed by definitive erythropoiesis in zebrafish sequencing of the locus, produced mCherry+ cells in the CHT and We evaluated the separate and combined contributions of the two in the kidney in 10% (n=15/150) of embryos at 3 dpf (Fig. S3B) and at Hemogen promoters, including CNE1 or CNE2, to the observed a lower frequency in circulating RBC (n=3/150, data not shown). tissue-expression profiles by injecting wild-type embryos with one To characterize hematopoietic cell lineages that express of three Tg(Hemgn:mCherry,myl7:EGFP) reporter constructs in Hemogen, the P1 reporter plasmid was injected into embryos of which mCherry expression was driven: (1) by the dual promoter Tg(CD41:EGFP)Ia2Tg or Tg(Lcr:EGFP)cz3325Tg zebrafish, which (P1); (2) by a 2 kb fragment (P2) containing the distal promoter have been used to track hematopoietic progenitors (Lin et al., 2005) including CNE1; or (3) by a 188 bp fragment (P3) containing the and primitive and definitive erythrocytes (Ganis et al., 2012), proximal promoter including CNE2 (Fig. 6A). Transgenic embryos respectively. We did not observe mCherry expression in the AGM, were screened for EGFP+ hearts, and mCherry transcription was in the thymus, or in CD41+ HSPCs colonizing the thymus or confirmed by RT-PCR and sequencing. pronephros (Bertrand et al., 2008). However, the reporter was mCherry fluorescence was examined in four cell types: (1) strongly expressed in a subset of LCR+ erythroid and CD41+ erythroid progenitors in the ICM at 1 dpf, (2) primitive erythrocytes myeloid-biased progenitors in the CHT (Fig. 6E,F), a tissue that in the peripheral blood at 3 dpf, (3) erythroid progenitors in the CHT supports myelopoiesis (Gekas and Graf, 2013; Medvinsky et al., at 3 dpf and (4) renal cells of the kidney tubules at 3 dpf. Fig. 6J 2011). This lends support to previous findings that Hemogen is a shows that the dual promoter (P1) supported strong expression of marker and promoter of myeloerythroid, but not lymphoid, lineages the mCherry reporter in erythroid cells of the ICM and peripheral (Li et al., 2007; Lu et al., 2001). Maturing mCherry+ primitive blood (RBC), in the CHT and in renal cells of the kidney. By progenitors peaked in brightness just prior to leaving the caudal contrast, the distal promoter (P2 construct) containing CNE1 failed plexus and entering circulation at 72 hpf (observed by time-lapse to drive reporter expression in these tissues. Finally, the proximal imaging; data not shown). However, mature definitive erythrocytes promoter (P3 construct) containing CNE2 alone produced strong expressed little mCherry in adult transgenics (Fig. 6G), which expression of the reporter in the CHT and in kidney cells but was not Biology Open

6 RESEARCH ARTICLE Biology Open (2018) 7, bio035576. doi:10.1242/bio.035576

At 24 hpf, 61% of morphants were anemic compared to 35% of uninjected zebrafish (Fig. 7A,B). Red cell levels were restored to wild-type by co-injection of the MO with 500 pg of synthetic zebrafish Hemogen mRNA containing silent mutations in the MO target site. Both the uninjected and rescue treatments differed significantly from the MO treatment (ANOVA, Tukey post hoc test, P<0.001, Fig. 7A). We used Tg(Lcr:EGFP)cz3325Tg zebrafish to visualize the red blood cell population in Hem1-treated morphants from 0 to 6 dpf. Control embryos were injected with a 5 bp mismatch MO (Hem1mm) or were uninjected. At 20 hpf, EGFP+ erythrocytes appeared to be reduced in the ICM/PBI of 75% of Hem1 morphants (n=56) but not in mismatch or uninjected control embryos (n=14 and 63, respectively) (Fig. 7C-E). At 2 dpf, morphant embryos had few erythrocytes in circulation compared to controls (Fig. 7F-H, Movie 2). Using quantitative in vivo flow analysis (Fig. 7I), we found that morphant embryos at 3-5 dpf had fewer than 50% of the circulating EGFP+ erythrocytes as the uninjected and Hem1mm-injected controls, whereas the controls did not differ statistically from each other (ANOVA, Tukey- Kramer post hoc test, P<0.05).

A conserved C-terminal domain in Hemogen is required for hematopoiesis and prevents apoptosis in embryonic tissues The function of the putative C-terminal transactivation domain of zebrafish Hemogen was investigated using CRISPR/Cas9 mutagenesis. We generated zebrafish lines with mutations in the conserved region near the end of the third coding exon of Hemogen, immediately downstream of the TAD motif (Fig. 8A-D; Fig. S1). Founders (F0) were out-crossed to wild-type TU zebrafish and mutant alleles were genotyped in the F1 generation by high resolution melting analysis and by sequencing the locus (Fig. 8E; Fig. S1). One line, Hemgnnuz2, had a 5 bp deletion (Δ5) that produced a frameshift mutation, thereby introducing a premature stop codon (Fig. 8E; Fig. S1). PolyA-tailed transcripts of the Δ5 allele were detected at equivalent steady-state levels relative to Fig. 5. Gata1 binds distal and proximal promoter elements to regulate the wild-type allele in peripheral blood from individual adult Hemogen expression in zebrafish. (A) Gata1 ChIP-sequencing showing heterozygotes (Fig. 8F). Western blot analysis revealed, however, enriched binding of Gata1 at CNE1 and CNE2 (C1 and C2, red lines) in the Hemogen promoter in adult zebrafish red blood cells (Yang et al., 2016). that truncated Hemogen protein was almost undetectable in DNase-sequencing and ATAC-sequencing showing co-localization of the peripheral blood from single heterozygous adults (data not active chromatin regions (Yang et al., 2016). (B) Hemogen expression by shown) and in pooled 33 hpf embryos from a heterozygous in- WISH of wild-type (n=16/21) and (C) homozygous mutant (n=5/21) siblings cross (Fig. 8G). Therefore, if the truncated Hemgnnuz2 transcripts − (33 hpf) from in-crossed Gata1+/ vltm651 mutants. Insets show βe1-globin were translated, then the protein must have been rapidly degraded. expression in mutant (n=4/10) and wild-type (n=6/10) siblings. Scale bar: The second line, Hemgnnuz4, contained an in-frame 12 bp deletion 250 µm (C). (Δ12), which deleted an acidic cluster (EEED) in the last repeat that is conserved in teleost species (Fig. S1). In contrast to Δ5 mutants, active in cells of the ICM and peripheral blood. Together, these Hemogen protein was detected in the blood of homozygous Δ12 results indicate that the proximal promoter containing CNE2 is adults by western blot (data not shown). necessary and sufficient to drive expression in definitive To evaluate the effects of the mutant Hemogen alleles on hematopoiesis and in the kidney, whereas the full 2.2 kb sequence erythropoiesis during development, we examined embryos from including both promoters and CNEs is required in primitive mutant crosses by microscopy and genotyped them between 20 and erythropoiesis. 48 hpf (Fig. 8A-C); mutant genotypes were recovered near the expected Mendelian ratios (Fig. S5A), but homozygous Δ5 Morpholino knockdown of Hemogen protein expression hemgnnuz2 mutants could not be raised to adulthood. To classify partially disrupts erythropoiesis in zebrafish the mutants, we assessed the relative numbers of blood cells and To perturb Hemogen function in zebrafish, we first injected wild- relative concentrations of hemoglobin beginning at 2 dpf (Ransom type zebrafish embryos at the one-cell stage with an antisense et al., 1996). Embryos from a heterozygous in-cross were scored for morpholino oligonucleotide (MO), Hem-1, targeted to the hypochromic blood (paler blood) and decreased numbers of translation start codon of the Hemogen transcript (Hem1). MO circulating cells on the yolk sac and in the vasculature. treatment significantly reduced Hemogen protein levels by 19% at Erythrocyte levels were reduced to about 25-75% of normal levels 33 hpf (Student’s t-test, P<0.05, Fig. S4A,B) and steady-state levels in frameshift Hemgnnuz2/+ mutants (n=8) at 24 hpf compared to of βe1-globin mRNA at 3 dpf (Student’s t-test, P<0.05, Fig. S4C). wild-type siblings (n=7) (Fig. 8C). At 48 hpf, 59% of heterozygous Biology Open

7 RESEARCH ARTICLE Biology Open (2018) 7, bio035576. doi:10.1242/bio.035576

Fig. 6. Promoter elements have distinct roles in driving hematopoietic, renal and testicular expression of Hemogen in transgenic Tg(Hemgn:mCherry) zebrafish. (A) Schematic of the zebrafish Hemogen gene. CNEs, black; coding exons, white; transcription initiation sites, bent arrows. Three Tg(Hemgn:mCherry, myl7:EGFP) transgenes driven by portions of the Hemogen promoter were transfected into one-cell TU embryos by Tol2 transposase-mediated insertion. Numbers indicate length of promoter elements and arrows show gene direction. (B) 20 hpf. P1 transgene expression in the peripheral blood island (PBI). (C) 72 hpf. P1 transgene expression in the pronephric ducts (PD). (D) 5 dpf. P1 transgene expression in the proximal convoluted tubule (PCT). (E,F) 72 hpf. Co-localization of mCherry and EGFP in progenitors in the CHT of Tg(Hemgn-P1:mCherry,Lcr:GFP) or Tg(Hemgn-P1:mCherry,CD41:EGFP) zebrafish. (G) Transgene expression in mature erythrocytes from adult zebrafish. (H) Transgene expression in adult head kidney (HK), trunk kidney (TK) and tail kidney (T) near the EGFP+ heart (H). (I) Transgene expression in adult Sertoli cells (Se) that surround the seminiferous tubules (ST). (J) Proportion of embryos expressing transgenes P1, P2 or P3 in ICM, kidney, CHT and circulating primitive erythrocytes (RBC). Scale bars: 100 µm (B, D-F,I); 500 µm (C,H); 25 µm (G).

(n=49) and 50% of homozygous (n=12) Hemgnnuz2 mutants had growth was significant in homozygous Δ12 Hemgnnuz4 adult reduced numbers of circulating erythrocytes (Fig. 8H, Movie 3) and mutants (Student’s t-test, P=0.04, n=3; Fig. S5D,E). homozygotes could be distinguished by their more severe anemia. Comparable numbers of anemic individuals were observed for DISCUSSION heterozygotes and homozygotes of the Δ12 Hemgnnuz4 allele: 64% The zebrafish is a compelling model for understanding the (n=25) and 60% (n=10), respectively (Fig. 8H). In all cases, the pleiotropic functions of Hemogen in the context of vertebrate proportion of anemic mutant embryos was significantly different development. Our results show that zebrafish Hemogen is from that for wild-type (*P≤0.05, **P≤0.005, Chi square). considerably smaller than its human ortholog, a distinction true Erythrocyte levels in adult mutants were partially suppressed in for teleost and mammalian Hemogens in general. Hemogen is heterozygotes. Hemgnnuz2/+ and Hemgnnuz4/+ adults gave average expressed in multiple zebrafish tissues from the early embryo to erythrocyte counts of 2.2±1.0×106 cells µl−1 and 2.1 the adult under the control of at least two promoters. Both ±0.8×106 cells µl−1, respectively, whereas wild-type zebrafish had primitive and definitive erythropoiesis are affected by depletion of 4.3±1.0×106 cells µl−1 (Fig. 8J,K). Homozygous Δ12 Hemgnnuz4 Hemogen and by targeted mutation of a putative, C-terminal TAD. gave average erythrocyte counts of 2.2±1.2×106 cells µl−1 The transgenic and mutant zebrafish lines that we have generated (Fig. 8K). Taken together, the erythroid defects of embryonic and will contribute to a mechanistic understanding of this important adult zebrafish carrying the CRISPR-generated mutant alleles transcription factor. support the conclusion that the conserved C-terminus of Hemogen functions as a TAD, but the mechanism of action of these mutations Hemogen – small or large, it’s built of related modules and remains to be determined. has a conserved role in erythropoiesis Both the Δ5 and Δ12 mutant Hemogen alleles also caused mild to We show that the divergent Hemogens of zebrafish and human are severe developmental defects in the nototchord and the trunk of largely, but not entirely, built of 21-25 residue repeats; the number heterozygotes and homozygotes (Fig. 8A,B; Fig. S5B). Embryos of repeats largely determines protein size. The repeat consensus had kinked notochords and exhibited increased cellular refractility sequences are distinct, but they appear to have evolved from an 8-10 consistent with apoptotic cell death. Elevated apoptotic cell death amino acid core motif (Fig. S2). Although all repeats are acidic was apparent in Hemgnnuz2/+ mutants as detected by staining with (Fig. S2), the terminal repeat of each Hemogen is particularly so Acridine Orange (Fig. S5C). Apoptosis occurred throughout the (>38% Asp and Glu for zebrafish, >29% for human), and these embryo, including sites of embryonic hematopoiesis. Nevertheless, repeats contain TAD motifs. Together, these features suggest that viable heterozygotes for both alleles could be raised to adulthood; Hemogens possess flexible, intrinsically disordered TADs, as is they were slightly smaller than wild-type siblings (Fig. 8I). Impaired true of many transcription factors (e.g. p53, HIF-1α, NF-κB, etc.). Biology Open

8 RESEARCH ARTICLE Biology Open (2018) 7, bio035576. doi:10.1242/bio.035576

Fig. 7. Morpholino targeting of Hemogen inhibits erythropoiesis in embryonic zebrafish. Embryos were injected with 2 to 4 ng antisense MO targeted to the first 25 coding nucleotides of Hemogen. (A-B) O-dianisidine staining of erythrocytes was decreased in morphants (MO) relative to wild-type embryos (WT) or embryos rescued with 500 pg synthetic Hemgn mRNA (zHem) at 24 hpf. (ANOVA, Tukey post hoc test, P<0.001). (C-E) Live wild-type (C), Hem1 MO-injected (D) and Hem1 mm mismatch MO-injected (E) Tg(Lcr:EGFP)cz3325Tg embryos at 20 hpf. Morphants showed decreased EGFP expression in the ICM compared to the wild-type and mismatch MO controls. (F-H) Live wild-type (F), Hem1 MO-injected (G) and Hem1 mm MO- injected (H) embryos at 72 hpf. Morphant embryos have fewer EGFP+ cells in circulation compared to the two controls. The dorsal aortas of embryos (insets above F-H) were magnified 20× to permit quantitation of EGFP+ erythrocytes. Background red (D,G) and green (E,H) fluorescence was generated by the fluorescent labels on the MOs. (I) In vivo flow quantitation of EGFP+ erythrocyte concentrations between 3 and 6 dpf in Hem1-injected (n=9,7,7,7), Hem1mm-injected (n=13,14,11,11) and uninjected (n=5,10,10,9) embryos. Data shown as means±s.e.m. (*P≤0.05, **P≤0.001, ANOVA, Tukey-Kramer post hoc test). Arrowheads show notochord kinking. Scale bars: 500 µm (A-F); 100 µm (inset).

Hemogen interacts with a variety of proteins to stimulate the transcription of genes involved in terminal erythroid differentiation and other processes. In humans, Hemogen contributes to transcription of erythroid genes in part by recruiting P300 to acetylate and activate Gata1 (Zheng et al., 2014). Our results show that nonsense (Δ5) and deletion (Δ12) alleles of Hemogen vicinal to the zebrafish TAD motif cause significant reductions of erythrocyte levels in embryos and adults. The Δ12 allele may be hypomorphic, but we have not determined whether the protein that is expressed has reduced activity.

Hemogen – targeted mutation of the acidic C-terminus impairs erythropoiesis, but not completely Our CRISPR-generated zebrafish mutant lines show that nonsense (Δ5) and deletion (Δ12) alleles of Hemogen caused a decrease in erythrocyte levels in embryos and adults. However, these phenotypes were incompletely penetrant; in both heterozygous and homozygous Hemogen mutants the proportion of anemic embryos was 50-65%, compared to 20% for wild types. If Hemogen were essential for erythropoiesis, one would anticipate an erythroid- null phenotype for homozygous mutants, as observed for the Gata1 mutant, vlad tepesm651 (Lyons et al., 2002). Rather, the Hemogen phenotype resembles the variable reduction of red cells in zebrafish zinfandel (zinte207) mutants that harbor a mutation in a regulatory region at the globin locus (Ransom et al., 1996), a known target of both Hemogen and Gata1 transcription factors (Zheng et al., 2014). Loss of Hemogen in zebrafish contributes to decreased expression of Embryonic beta-globin (Fig. S4), which may explain the hypochromic state of Hemogen mutants. The most plausible explanation for the incomplete penetrance of anemia in Hemogen mutants is the phenomenon of genetic compensation, which may occur when genes are knocked out as opposed to knocked down (El-Brolosy and Stainier, 2017; Rossi et al., 2015). Although the mechanisms are poorly understood, genetic compensation entails changes in gene expression (e.g. upregulation of paralogous genes or functionally related genes) that at least partially offset the phenotype caused by the mutant protein. Compensation through elevated expression of other erythroid co- activators is an attractive possibility that might maintain erythrocyte production in Hemogen mutants. The functional loss of Hemogen The multivalent structure of Hemogen provides opportunities for could be mitigated by Gata1 homodimerization and/or by direct cooperative binding to single or multiple protein partners, including recruitment of CBP/P300, both of which enhance Gata1 activity

P300 (Zheng et al., 2014). (Ferreira et al., 2005; Nishikawa et al., 2003). Biology Open

9 RESEARCH ARTICLE Biology Open (2018) 7, bio035576. doi:10.1242/bio.035576

Fig. 8. CRISPR/Cas9 mutagenesis of the third exon of zebrafish Hemogen reduces primitive and definitive erythropoiesis. Embryos were injected with Cas9 mRNA and a guide RNA to establish lines with mutations in exon three of zebrafish Hemogen. (A) 20 hpf. Representative wild-type and mutant siblings with notochord defects (arrow). (B) 48 hpf. Mutant Δ12 embryos with an in-frame deletion showing kinked notochords (arrow). (C) 24 hpf. Wild-type and Δ5/+ mutant embryos stained with diaminofluorene. Production of erythrocytes was reduced in heterozygotes. (D) Schematic of CRISPR/Cas9 target in the third exon (red arrowhead) of zebrafish Hemogen. (E) Sequences of founder mutations aligned at the CRISPR target site: Δ5 (Hemgnnuz2); Δ12 (Hemgnnuz4). The sequence traces show the Δ5 and Δ12 mutant alleles. PAM, blue and underlined; Δ, deletions (highlighted in red). (F) Relative expression of wild-type and Δ5 transcripts in blood from single adult, heterozygous Hemgnnuz2/+ mutants determined by qRT-PCR with allele specific primers. Three biological replicates were normalized to β-actin. Error bars represent the standard deviation. (G) Western blot of Hemogen in pooled 33 hpf wild-type embryos or pooled embryos from a Δ5 Hemgnnuz2/+ heterozygous in-cross. We calculated that the protein would run 6.5 kDa above its molecular weight at 28.5 kDa because of its high acidic composition (Guan et al., 2015). Arrows show the calculated sizes of wild-type and truncated alleles. (H) Proportion of genotyped mutants and wild-type sibling embryos at 2 dpf that were anemic (black) or phenotypically normal (white) (*P≤0.05, **P≤0.005, Chi square). (I) Wild-type and mutant zebrafish heterozygous for the Δ5 and Δ12 alleles. (J) Red blood cells from adult Hemgnnuz2/+ mutant zebrafish and wild-type siblings. (K) Erythrocyte counts in adult heterozygous Hemgnnuz2 (Δ5, n=12), heterozygous Hemgnnuz4 (Δ12, n=4) mutants, homozygous Hemgnnuz4 (Δ12, n=2) mutants and wild- type (n=9) siblings (*P≤0.05, ANOVA, Tukey post hoc test). Scale bars: 500 µm (A-C); 50 mm (I); 20 µm (J).

Similar design and regulation of Hemogen and Gata1 genes at the midbrain-hindbrain boundary (Volkmann et al., 2008). Comparison of the expression of Hemogen and of Gata1 Interestingly, both Hemogen and Gata1 genes possess throughout zebrafish development reveals a remarkable degree hematopoietic- and testis-specific promoters (Wakabayashi et al., of overlap in tissue and cellular specificity. For example, Gata1 2003). The temporal and spatial co-incidence of Hemogen and mRNA appears in cells of the LPM at the two-somite stage Gata1 expression almost certainly results from their similar (Detrich et al., 1995), immediately prior to the onset of Hemogen regulatory architectures and also through regulatory crosstalk. Our expression at ten somites. Furthermore, Hemogen and Gata1 results and studies conducted by others (Ding et al., 2010; Yang are co-expressed in primitive erythrocytes and definitive et al., 2006; Zheng et al., 2014) indicate that reciprocal hematopoietic progenitors (Ferreira et al., 2005; Lu et al., 2001), transcriptional activation of Hemogen and Gata1 may form a in Sertoli cells (Nakata et al., 2013; Wakabayashi et al., 2003) and positive feedback loop that drives erythropoiesis. Biology Open

10 RESEARCH ARTICLE Biology Open (2018) 7, bio035576. doi:10.1242/bio.035576

Strikingly, the two CNEs of Hemogen are organized like, and have et al., 2005) and the Conserved Domain Database (CDD) (Marchler-Bauer the same functions as, the distal and proximal enhancers of the Gata1 et al., 2015). Peptide repeats were predicted with RADAR (Heger and Holm, gene (McDevitt et al., 1997; Onodera et al., 1997; Suzuki et al., 2000). The 9aaTAD Prediction Tool was first used to predict transactivation domain (TAD) motifs, starting with low stringency DFx repeats (Piskacek 2009). The proximal Gata1 promoter functions exclusively in φφ φ φ φφ φ definitive erythropoiesis (McDevitt et al., 1997), as does CNE2 of et al., 2016). These were then culled by xx or xx criteria, where is a bulky hydrophobic motif (Dyson and Wright, 2016). We refer to the latter zebrafish Hemogen. In contrast, transcription of Gata1 in primitive five amino acid consensus sequences as ‘TAD motifs’, in contrast to larger, erythrocytes requires both the proximal promoter and a distal functionally defined ‘transactivation domains’ (TADs). Ab initio tertiary enhancer comparable to Hemogen CNE1 (McDevitt et al., 1997). structure models were created for zebrafish and human Hemogen proteins Fig. S6 presents a model for the transition from primitive to definitive with I-Tasser (Yang et al., 2015) based on the X-ray structure for the hematopoiesis based on chromatin looping at the Hemogen locus. We secretory component of Immunoglobulin A (PDB:3chnS), which was the propose that the transition from primitive to definitive erythropoiesis best of ten predicted structural templates determined by LOMETS (Wu and involves a switch from a loop conformation to a linear conformation, Zhang, 2007). The 3D models were superimposed using TM-align (Zhang mediated by the Gata1/Ldb1-complex at erythroid transcription and Skolnick, 2005) and Geneious version R10 (Kearse et al., 2012). factories (Osborne et al., 2004; Schoenfelder et al., 2010). This model may also apply to the Gata1 enhancer, which is another known target MO knockdown of Hemogen in zebrafish and rescue of the of the Ldb1-complex (Love et al., 2014). The zebrafish lines morphant phenotype produced in this study may help clarify the cell-specific Hemogen The antisense MO Hem1 (5′-TCTCTTTCTCCAACGGGTCTTCCAT-3′), expression profile driven by different Gata1-containing complexes which targets the first 25 base pairs of the zebrafish Hemogen open and the functions of Hemogen in different cell types. reading frame, was designed according to the manufacturer’s instructions (Gene Tools LLC, Philomath, USA). The control MO (Hem1mm; MATERIALS AND METHODS 5′-TCTgTTTgTCCAtCGGcTCTTCgAT-3′) targeted the same sequence Fish husbandry but contained five mismatched bases to prevent efficient binding to Wild-type (SAT, AB, TU) zebrafish (Danio rerio), the transgenic lines Hemogen mRNA. MOs were labeled with lissamine or fluorescein so that Tg(Lcr:EGFP)cz3325Tg (Ganis et al., 2012) and Tg(CD41:EGFP)Ia2Tg the quality of injections could be monitored by fluorescence microscopy. (Traver et al., 2003) and the mutant vlad tepesm651 (Lyons et al., 2002) MOs were injected (2-8 ng) into embryos at the single-cell stage using a were all generously provided by Dr Leonard I. Zon (Howard Hughes PLI-100 Picoinjector (Medical Systems Corporation, Greenvale, USA; Medical Institute and Harvard Medical School, Boston). Animal procedures 65-0001) and a micromanipulator (Narishige, Amityville, USA; MN-151). were carried out in full accordance with established standards set forth in the Injected embryos were sampled from 0 to 6 dpf for subsequent analyses. Guide for the Care and Use of Laboratory Animals (8th Edition). The Rescue of the morphant phenotype was tested by co-injection of the animal care and use protocol for live zebrafish embryos was reviewed and Hem1 MO with 500 pg synthetic zebrafish Hemogen mRNA transcribed approved by Northeastern University’s Institutional Animal Care and Use from a zebrafish Hemogen cDNA cloned into pGem-T Easy (Promega). Committee (Protocol No. 15-0207R). The animal care and use program at Primers (Table S1) introduced five silent mutations within the MO target Northeastern University has been continuously accredited by AAALAC Int. site. The clone was digested with Spe1 and mRNA was transcribed, capped since 22 July 1987 and maintains the Public Health Service Policy and polyadenylated in vitro using the mMessage T7 kit (Ambion; AM1340) Assurance Number A3155-01. and the Poly(A) Tailing Kit (Ambion; AM1350). mRNA was purified with the MEGAclear kit (Ambion). Cloning and sequence analysis of zebrafish Hemogen cDNAs Total RNA was isolated from wild-type AB zebrafish embryos and adult In situ hybridization tissues (kidney, blood, brain, ovary, intestine) using TRI reagent (Sigma- The spatial and temporal patterns of expression of selected genes were Aldrich; T9424) and the Ribopure Kit (Ambion, Foster City, USA; analyzed by whole-mount in situ hybridization (WISH) of zebrafish AM1924). Total cDNA was produced from mRNA using M-MuLV reverse embryos following standard protocols (Jacobs et al., 2011). These methods transcriptase [New England Biolabs (NEB), Ipswich, USA; M0253S] and were adapted to evaluate Hemogen expression in tissues, peripheral blood an oligo(dT)23 primer. Hemogen cDNA was amplified by PCR from total smears and pronephric kidney prints prepared from euthanized adult fish cDNA with 1 µM primers (Table S1). The amplification program was 35 [200 mg l−1 tricaine methane sulfonate (MS222; Sigma-Aldrich, 886862)] cycles of 98°C for 10 s, 57°C for 10 s and 72°C for 30 s. PCR products were (Detrich and Yergeau, 2004; Gupta and Mullins, 2010). For sectioning, cloned into the pGEM-T Easy vector (Promega, Madison, USA; A1360), embryos and tissues were embedded in a solution containing 0.25 g gelatin, plasmids were transformed into 5-α competent cells (NEB; C2987H), 30 g albumin, 22 g sucrose, 2.5% glutaraldehyde (v/v) per 100 ml recombinant plasmids were id entified by blue/white screening and purified phosphate buffered saline (PBS). Sections were cut with a vibrating blade with the Wizard Plus SV Miniprep Kit (Promega; A1330) and inserts were microtome (Leica, Wetzlar, Germany; VT1000S). Digoxigenin-labeled sequenced by GENEWIZ, Inc. (Cambridge, USA). antisense and sense RNA probes were transcribed from zebrafish cDNA clones using the DIG RNA Labeling Kit (Roche Diagnostics, Indianapolis, Bioinformatic comparison of vertebrate Hemogen genes and USA; 11175025910). Hemogen proteins We utilized the murine gene nomenclature for comparing orthologs from Indirect immunofluorescence different vertebrate species. We used Blast+ (Altschul et al., 1990) to Zebrafish embryos were fixed in 4% paraformaldehyde (PFA) at 48 hpf. identify Hemogen in the zebrafish genome (assembly GRCz11) (Howe Embryos were incubated with 1:1000 rabbit anti-Hemogen primary et al., 2013). Chromosomal synteny comparisons were performed using the antibody (Aviva, San Diego, USA; ARP57794_P050) followed by 1:1000 Synteny Database with a sliding window of 200 genes (Catchen et al., 2009) goat anti-rabbit IgG Alexafluor 488 secondary antibody (Life Technologies; and Ensembl Genomes v74 (Kersey et al., 2016). Hemogen promoter A11034) as previously described (Westerfield, 2000). The specificity of the alignments were obtained from whole genome alignments for ten teleost Hemogen antibody was validated both by Clontech (Mountain View, USA) species (ENSEMBL v74) (Yates et al., 2016). Transcription factor binding and by our laboratory by western blotting of zebrafish protein extracts. motifs were predicted using the program ConTra with the default similarity matrix of 0.75 (Broos et al., 2011). Transcription start sites were predicted Hemoglobin staining using NNPP v2.2 with a score cutoff of 0.98 (Reese, 2001). To detect red blood cells in circulation, embryos were stained with Protein domains in zebrafish were identified using annotated human o-dianisidine (Iuchi and Yamamoto, 1983) or diaminofluorene (McGuckin

Hemogen (Yang et al., 2001), or they were predicted using HHpred (Soding et al., 2003). Biology Open

11 RESEARCH ARTICLE Biology Open (2018) 7, bio035576. doi:10.1242/bio.035576

Western blotting between 1 and 3 dpf. To confirm integration, the locus was PCR amplified Total embryonic protein was prepared for sodium dodecyl sulfate with internal and external primers (Table S1) and cloned into the pGem-T polyacrylamide gel electrophoresis (SDS-PAGE) from dechorionated, Easy vector for sequencing. 33 hpf embryos (n=80) by homogenization in lithium dodecyl sulfate Wild-type (TU) embryos were co-injected with a guide RNA (LDS) Bolt buffer (Life Technologies; B007) and NuPAGE reducing agent (150 ng µl−1) targeting exon 3, Cas9 mRNA (300 ng µl−1) and mCherry (Life Technologies; NP0009) using a pestle and microcentrifuge tube (USA mRNA (30 ng µl−1) to identify successful injections. Embryos were raised Scientific, Ocala, USA; 1415-5390). Samples were boiled for 3 min and and adults were tail-clipped for haplotyping by high-resolution melting centrifuged at top speed in a centrifuge for 2 min. Aliquots (15 µg) were analysis (HRMA) as previously described (Talbot and Amacher, 2014). electrophoresed on a 4-12% SDS polyacrylamide gel, and the separated PCR amplification was run using 1 µM primers (Table S1) with PowerUp proteins were transferred to a polyvinylidene difluoride (PVDF) membrane SYBR MasterMix (Applied Biosystems, Foster City, USA; A25742) on a with the iBlot system (Life Technologies; IB21001). Membranes were QuantStudio 3 Real-time PCR system (Thermo Fisher Scientific; A28137). blocked in maleic acid blocking buffer (2% Roche blocking reagent, 2% Founder mutants were outcrossed to TU fish. The offspring were raised and BSA, 0.2% heat treated goat serum, 0.1% Tween-20) for 1 h at room mutations were characterized by HRMA and sequencing of the locus. temperature and then incubated overnight at 4°C with 1:1000 rabbit anti- Hemogen (Aviva; ARP57794_P050) or with 1:1000 mouse anti-GAPDH Imaging of zebrafish embryos (Aviva; OAE00006) antibodies. Membranes were washed in TBST (Tris- Fixed embryos were mounted in 80% glycerol and imaged with a dissecting buffered saline and Tween 20) and incubated for 2 h with horseradish microscope (Nikon; SMZ-U) and a CCD digital camera (Diagnostic peroxidase (HRP)-conjugated goat anti-rabbit IgG (H&L) (Aviva; Instruments, Sterling Heights, USA; SPOT32). Live embryos were ASP00001) or HRP-conjugated goat anti-mouse IgG (H&L) (Aviva; embedded in 0.1% agarose in embryo medium (EB) with 0.01% tricaine OARA04973), respectively. Bound antibodies were detected with the and imaged with an epifluorescence-equipped microscope (Nikon; Eclipse Amersham ECL Western Blotting Analysis System (GE Healthcare; E800). Movies (0.01 s interval) and time-lapse images (1 min interval) were RPN2106) on CL-X Posure film (Thermo Fisher Scientific; 34091). obtained using a Photometrics Scientific CoolSNAP EZ camera and NIKON NIS-Elements AR 4.20 software. Methods for in vivo flow analyses Tol2 generation of Tg(Hemgn:mCherry) zebrafish were adapted to quantify fluorescently labeled red blood cells in MO- To identify the regulatory elements that drive Hemogen expression in injected Tg(Lcr:EGFP)cz3325Tg zebrafish (Schwerte et al., 2003; Zeng et al., zebrafish, three different Tg(Hemgn:mCherry) reporter plasmids were 2012). Briefly, 100 frame videos were taken set at a 500 μs exposure time created using Gateway Cloning Technology (Invitrogen; 11791020) with no delay. The field of view (20×) was centered on the dorsal aorta (Hartley et al., 2000). First, the proximal Hemogen promoter (∼2.2 kb) adjacent to the cloaca. The summed maximum intensity images of all frames was amplified from wild-type SAT zebrafish using 1 µM primers were used to create ‘casts’ of the dorsal aorta and the average volume was (Table S1). The promoter sequence spanned the upstream, non-coding calculated assuming cylindrical vasculature. EGFP+ cells were converted to region before, but not including, the Hemogen translation start codon. The binary objects (6.66 µm diameter, contrast 180) and counted within the promoter was cloned between KpnI/SpeI restriction sites in the p5e-MCS region of interest. vector (Tol2kit, http://tol2kit.genetics.utah.edu; #228) using the Tol2kit vector system (Kwan et al., 2007) to generate the entry clone, p5e-Hemgn-1. qRT-PCR The resulting plasmid was digested with NaeI/KpnI or NaeI/SpeI to remove RNA was purified from adult zebrafish tissues or 10-30 pooled embryos at 3 each of two conserved non-coding elements (CNE1 or CNE2) from the or 4 dpf in TriZol (Sigma-Aldrich; T9424) using the PureLink RNA promoter. Each new construct was blunt-ended with Q5 Hot Start High- purification Kit (Ambion). DNase treated RNA was reverse transcribed with Fidelity 2× Master Mix (NEB) and ligated with T4 DNA Ligase (NEB) to a polyT(23) primer using Protoscript II RT-PCR kit (NEB; M0368S). Target create p5e-Hemgn-2 and p5e-Hemgn-3. Each of the three entry clones were genes were amplified in triplicate from cDNA by qRT-PCR with 1 µM cloned in front of the mCherry gene within the pDestTol2CG2 destination primers (Table S1). Standard curves were generated to confirm primer vector (Tol2kit; #395). The pCS2FA-transposase clone (Tol2kit; #396) was efficiencies. Target gene expression was normalized to beta-actin for digested with PmeI, and Tol2 transposase mRNA was transcribed, capped comparison by the ΔΔCt method. Three or four biological replicates were and polyadenylated in vitro using the mMessage SP6 kit (Ambion; used for each treatment for statistical comparisons. AM1340) and the Poly(A) Tailing Kit (Ambion; AM1350). mRNA was −1 purified by precipitation using 2.5 M LiCl. Transposase mRNA (37 ng µl ) Statistical analyses and each of the Tg(Hemgn:mCherry,myl7:EGFP) expression clones Data were analyzed as means±s.e.m. or means±s.d. as noted. Statistical tests −1 (25 ng µl ) were co-injected into one-cell wild-type zebrafish embryos. applied to the results are provided with each experiment. Differences with a Founders were raised and out-crossed to wild-type TU zebrafish for two P-value≤0.05 were considered significant. generations. GenBank accession numbers CRISPR/Cas9 generation of transgenic and mutant zebrafish Zebrafish Hemgn isoform 1, JZ970258; zebrafish Hemgn isoform 2, Optimal targets for CRISPR-Cas9 mutagenesis were identified within the JZ970260; zebrafish Hemgn isoform 3, JZ970259; and zebrafish Hemgn first and third exons of zebrafish Hemogen using the program CHOPCHOP isoform 4, JZ970257. (Labun et al., 2016; Montague et al., 2014). The templates for multiple small guide RNAs were produced by a cloning-free method as previously Zebrafish ZFIN IDs described (Table S1) (Hruscha et al., 2013; Talbot and Amacher, 2014). Transgenic construct Tg(hemgn:mCherry,myl7:EGFP), ZDB-TGCONSTRCT- Guide RNAs were transcribed with the T7 MaxiScript Kit (Ambion; 170726-1; zebrafish line nuz1Tg, ZDB-ALT-170726-1; zebrafish line AM1312) and purified by LiCl precipitation. hemgnnuz2, ZDB-ALT-170726-2; zebrafish line hemgnnuz3, ZDB-ALT- A donor construct for homology directed repair was created containing 170726-3; zebrafish line hemgnnuz4, ZDB-ALT-170726-4. All transgenic the mCherry gene and polyadenylation signal flanked by 199 bp and 253 bp lines, with the exception of nuz3, are available through the Zebrafish homology arms that were PCR amplified from the sequence surrounding International Resource Center. exon 1 of Hemogen from wild-type AB zebrafish (Table S1). The homology arms and mCherry gene were PCR amplified with primers that added AvrII Acknowledgements and ClaI restriction sites, ligated and cloned into the pGem-T Easy vector ’ cz3325Tg We thank Dr Leonard Zon and Christian Lawrence at the Children s Hospital in (Promega). Tg(Lcr:EGFP) embryos were co-injected at the single- Boston for providing zebrafish and plasmids. We thank Dr John Postlethwait, Dr −1 cell stage with EcoRI linearized donor plasmid (25 ng µl ), two exon-1 Leonard Zon and Christopher Wells for helpful discussion. We thank Dr Johanna −1 −1 targeting guide RNAs (150 ng µl ) and Cas9 mRNA (300 ng µl ) Farkas and Carly Ching for their technical contributions. We thank Dr Leonard Zon,

(Trilink, San Diego, USA). Embryos were checked for fluorescence Dr Yi Zhou and colleagues at the Boston Children’s Hospital, Stem Cell and Biology Open

12 RESEARCH ARTICLE Biology Open (2018) 7, bio035576. doi:10.1242/bio.035576

Regenerative Biology Department, Harvard Medical School and Harvard University Ferreira, R., Ohneda, K., Yamamoto, M. and Philipsen, S. (2005). GATA1 for providing ATAC-seq, ChIP-seq and DNase I-seq datasets. function, a paradigm for transcription factors in hematopoiesis. Mol. Cell. Biol. 25, 1215-1227. Competing interests Forbes, S. A., Beare, D., Boutselakis, H., Bamford, S., Bindal, N., Tate, J., Cole, The authors declare no competing or financial interests. C. G., Ward, S., Dawson, E., Ponting, L. et al. (2017). COSMIC: somatic cancer genetics at high-resolution. Nucleic Acids Res. 45, D777-D783. Galloway, J. L. and Zon, L. I. (2003). Ontogeny of hematopoiesis: examining the Author contributions emergence of hematopoietic cells in the vertebrate embryo. Curr. Top. Dev. Biol. Conceptualization: M.J.P., S.K.P., J.G., C.A.H.A., H.W.D.; Methodology: M.J.P., 53, 139-158. S.K.P., J.G., C.A.H.A., J.L., H.W.D.; Formal analysis: M.J.P., S.K.P., J.G., C.A.H.A., Ganis, J. J., Hsia, N., Trompouki, E., de Jong, J. L. O., DiBiase, A., Lambert, J.L., H.W.D.; Investigation: M.J.P., S.K.P., C.A.H.A., J.L., H.W.D.; Resources: J. S., Jia, Z., Sabo, P. J., Weaver, M., Sandstrom, R. et al. (2012). Zebrafish M.J.P., H.W.D.; Data curation: M.J.P., S.K.P., J.G., C.A.H.A., J.L.; Writing - original globin switching occurs in two developmental stages and is controlled by the LCR. draft: M.J.P., H.W.D.; Writing - review & editing: M.J.P., H.W.D.; Visualization: M.J.P., Dev. Biol. 366, 185-194. H.W.D.; Supervision: M.J.P., S.K.P., H.W.D.; Project administration: H.W.D.; Gardiner, M. R., Gongora, M. M., Grimmond, S. M. and Perkins, A. C. (2007). A Funding acquisition: M.J.P., H.W.D. global role for zebrafish klf4 in embryonic erythropoiesis. Mech. Dev. 124, 762-774. Funding Gekas, C. and Graf, T. (2013). CD41 expression marks myeloid-biased adult hematopoietic stem cells and increases with age. Blood 121, 4463-4472. This research was supported by a Graduate Research Grant from the College of Guan, Y., Zhu, Q., Huang, D., Zhao, S., Jan Lo, L. and Peng, J. (2015). An Sciences and the Office of the Vice Provost of Graduate Studies at Northeastern equation to estimate the difference between theoretically predicted and SDS University, awarded to M.J.P. and by National Science Foundation grants PLR- PAGE-displayed molecular weights for an acidic peptide. Sci. Rep. 5, 13370. 1247510 and PLR-1444167 awarded to H.W.D. This is contribution number 380 Gupta, T. and Mullins, M. C. (2010). Dissection of organs from the adult zebrafish. from the Northeastern University Marine Science Center. J. Vis. Exp.: JoVE 37, e1717. Hartley, J. L., Temple, G. F. and Brasch, M. A. (2000). DNA cloning using in vitro Supplementary information site-specific recombination. Genome research 10, 1788-1795. Supplementary information available online at Heger, A. and Holm, L. (2000). Rapid automatic detection and alignment of repeats http://bio.biologists.org/lookup/doi/10.1242/bio.035576.supplemental in protein sequences. Proteins 41, 224-237. Hossain, M. S., Larsson, A., Scherbak, N., Olsson, P.-E. and Orban, L. (2008). References Zebrafish androgen receptor: isolation, molecular, and biochemical Altschul, S. F., Gish, W., Miller, W., Myers, E. W. and Lipman, D. J. (1990). Basic characterization. Biol. Reprod. 78, 361-369. Howe, K., Clark, M. D., Torroja, C. F., Torrance, J., Berthelot, C., Muffato, M., local alignment search tool. J. Mol. Biol. 215, 403-410. Collins, J. E., Humphray, S., McLaren, K., Matthews, L. et al. (2013). The An, L.-L., Li, G., Wu, K.-F., Ma, X.-T., Zheng, G.-G., Qiu, L.-G. and Song, Y.-H. zebrafish reference genome sequence and its relationship to the . (2005). High expression of EDAG and its significance in AML. Leukemia 19, Nature 496, 498-503. 1499-1502. Hruscha, A., Krawitz, P., Rechenberg, A., Heinrich, V., Hecht, J., Haass, C. and Bertrand, J. Y., Kim, A. D., Violette, E. P., Stachura, D. L., Cisson, J. L. and Schmid, B. (2013). Efficient CRISPR/Cas9 genome editing with low off-target Traver, D. (2007). Definitive hematopoiesis initiates through a committed effects in zebrafish. Development 140, 4982-4987. erythromyeloid progenitor in the zebrafish embryo. Development 134, 4147-4156. Hubank, M. and Schatz, D. G. (1999). cDNA representational difference analysis: a Bertrand, J. Y., Kim, A. D., Teng, S. and Traver, D. (2008). CD41+ cmyb+ sensitive and flexible method for identification of differentially expressed genes. precursors colonize the zebrafish pronephros by a novel migration route to initiate Methods Enzymol. 303, 325-349. adult hematopoiesis. Development 135, 1853-1862. Iuchi, I. and Yamamoto, M. (1983). Erythropoiesis in the developing rainbow trout, Broos, S., Hulpiau, P., Galle, J., Hooghe, B., Van Roy, F. and De Bleser, P. Salmo gairdneri irideus: histochemical and immunochemical detection of (2011). ConTra v2: a tool to identify transcription factor binding sites across erythropoietic organs. J. Exp. Zool. 226, 409-417. species, update 2011. Nucleic Acids Res. 39, W74-W78. Jacobs, N. L., Albertson, R. C. and Wiles, J. R. (2011). Using whole mount in situ Catchen, J. M., Conery, J. S. and Postlethwait, J. H. (2009). Automated hybridization to link molecular and organismal biology. J. Vis. Exp.: JoVE 49, e2533. identification of conserved synteny after whole-genome duplication. Genome Jiang, J., Yu, H., Shou, Y., Neale, G., Zhou, S., Lu, T. and Sorrentino, B. P. Res. 19, 1497-1505. (2010). Hemgn is a direct transcriptional target of HOXB4 and induces expansion Chen, D.-L., Hu, Z.-Q., Zheng, X.-F., Wang, X.-Y., Xu, Y.-Z., Li, W.-Q., Fang, H.-S., of murine myeloid progenitor cells. Blood 116, 711-719. Kan, L. and Wang, S.-Y. (2016). EDAG-1 promotes proliferation and invasion of Jin, H., Sood, R., Xu, J., Zhen, F., English, M. A., Liu, P. P. and Wen, Z. (2009). human thyroid cancer cells by activating MAPK/Erk and AKT signal pathways. Definitive hematopoietic stem/progenitor cells manifest distinct differentiation Cancer Biol. Ther. 17, 414-421. output in the zebrafish VDA and PBI. Development 136, 647-654. Cocca, E., Ratnayake-Lecamwasam, M., Parker, S. K., Camardella, L., Kearse, M., Moir, R., Wilson, A., Stones-Havas, S., Cheung, M., Sturrock, S., Ciaramella, M., di Prisco, G. and Detrich, H. W. (1995). Genomic remnants of Buxton, S., Cooper, A., Markowitz, S., Duran, C. et al. (2012). Geneious Basic: alpha-globin genes in the hemoglobinless antarctic icefishes. Proc. Natl. Acad. an integrated and extendable desktop software platform for the organization and Sci. USA 92, 1817-1821. analysis of sequence data. Bioinformatics 28, 1647-1649. Davidson, A. J. and Zon, L. I. (2004). The ‘definitive’ (and ‘primitive’) guide to Kersey, P. J., Allen, J. E., Armean, I., Boddu, S., Bolt, B. J., Carvalho-Silva, D., zebrafish hematopoiesis. Oncogene 23, 7233-7246. Christensen, M., Davis, P., Falin, L. J., Grabmueller, C. et al. (2016). Ensembl de Jong, J. L. and Zon, L. I. (2005). Use of the zebrafish to study primitive and Genomes 2016: more genomes, more complexity. Nucleic Acids Res. 44, definitive hematopoiesis. Annu. Rev. Genet. 39, 481-501. D574-D580. ̈ ̈ Detrich, H. W. and Yergeau, D. A. (2004). Comparative genomics in erythropoietic Kruger, A., Ellerstrom, C., Lundmark, C., Christersson, C. and Wurtz, T. (2002). gene discovery: synergisms between the Antarctic icefishes and the zebrafish. In RP59, a marker for osteoblast recruitment, is also detected in primitive Methods in Cell Biology, The Zebrafish, 2nd edition: Genetics, Genomics, and mesenchymal cells, erythroid cells, and megakaryocytes. Dev. Dyn. 223,414-418. Krüger, A., Somogyi, E., Christersson, C., Lundmark, C., Hultenby, K. and Informatics, Vol. 77 (ed. H. W. Detrich, M. Westerfield and L. I. Zon), pp. 475-503. Wurtz, T. (2005). Rat enamel contains RP59: a new context for a protein from San Diego: Elsevier Academic Press. osteogenic and haematopoietic precursor cells. Cell Tissue Res. 320, 141-148. Detrich, H. W., III, Kieran, M. W., Chan, F. Y., Barone, L. M., Yee, K., Rundstadler, Kwan, K. M., Fujimoto, E., Grabher, C., Mangum, B. D., Hardy, M. E., Campbell, J. A., Pratt, S., Ransom, D. and Zon, L. I. (1995). Intraembryonic hematopoietic D. S., Parant, J. M., Yost, H. J., Kanki, J. P. and Chien, C.-B. (2007). The Tol2kit: cell migration during vertebrate development. Proc. Natl. Acad. Sci. U.S.A. 92, a multisite gateway-based construction kit for Tol2 transposon transgenesis 10713-10717. constructs. Dev. Dyn. 236, 3088-3099. Ding, Y.-L., Xu, C.-W., Wang, Z.-D., Zhan, Y.-Q., Li, W., Xu, W.-X., Yu, M., Ge, C.- Labun, K., Montague, T. G., Gagnon, J. A., Thyme, S. B. and Valen, E. (2016). H., Li, C.-Y. and Yang, X.-M. (2010). Over-expression of EDAG in the myeloid cell CHOPCHOP v2: a web tool for the next generation of CRISPR genome line 32D: induction of GATA-1 expression and erythroid/megakaryocytic engineering. Nucleic Acids Res. 44, W272-W276. phenotype. J. Cell. Biochem. 110, 866-874. Li, C. Y., Zhan, Y. Q., Xu, C. W., Xu, W. X., Wang, S. Y., Lv, J., Zhou, Y., Yue, P. B., Dyson, H. J. and Wright, P. E. (2005). Intrinsically unstructured proteins and their Chen, B. and Yang, X. M. (2004). EDAG regulates the proliferation and functions. Nat. Rev. Mol. Cell Biol. 6, 197-208. differentiation of hematopoietic cells and resists cell apoptosis through the Dyson, H. J. and Wright, P. E. (2016). Role of intrinsic protein disorder in the activation of nuclear factor-kappa B. Cell Death Differ. 11, 1299-1308. function and interactions of the transcriptional coactivators CREB-binding Protein Li, C.-Y., Zhan, Y.-Q., Li, W., Xu, C.-W., Xu, W.-X., Yu, D.-H., Peng, R.-Y., Cui, Y.- (CBP) and p300. J. Biol. Chem. 291, 6714-6722. F., Yang, X., Hou, N. et al. (2007). Overexpression of a hematopoietic El-Brolosy, M. A. and Stainier, D. Y. R. (2017). Genetic compensation: a transcriptional regulator EDAG induces myelopoiesis and suppresses

phenomenon in search of mechanisms. PLoS Genet. 13, e1006780. lymphopoiesis in transgenic mice. Leukemia 21, 2277-2286. Biology Open

13 RESEARCH ARTICLE Biology Open (2018) 7, bio035576. doi:10.1242/bio.035576

Liao, E. C., Paw, B. H., Peters, L. L., Zapata, A., Pratt, S. J., Do, C. P., Lieschke, Schwerte, T., Uberbacher, D. and Pelster, B. (2003). Non-invasive imaging of G. and Zon, L. I. (2000). Hereditary spherocytosis in zebrafish riesling illustrates blood cell concentration and blood distribution in zebrafish Danio rerio incubated evolution of erythroid beta-spectrin structure, and function in red cell in hypoxic conditions in vivo. J. Exp. Biol. 206, 1299-1307. morphogenesis and membrane stability. Development 127, 5123-5132. Soding, J., Biegert, A. and Lupas, A. N. (2005). The HHpred interactive server for Lin, H.-F., Traver, D., Zhu, H., Dooley, K., Paw, B. H., Zon, L. I. and Handin, R. I. protein homology detection and structure prediction. Nucleic Acids Res. 33, (2005). Analysis of thrombocyte development in CD41-GFP transgenic zebrafish. W244-W248. Blood 106, 3803-3810. Soler, E., Andrieu-Soler, C., de Boer, E., Bryne, J. C., Thongjuea, S., Love, P. E., Warzecha, C. and Li, L. Q. (2014). Ldb1 complexes: the new master Stadhouders, R., Palstra, R.-J., Stevens, M., Kockx, C., van Ijcken, W. et al. regulators of erythroid gene transcription. Trends Genet. 30, 1-9. (2010). The genome-wide dynamics of the binding of Ldb1 complexes during Lu, J., Xu, W. X., Wang, S. Y., Zhan, Y. Q., Jiang, Y., Cai, W. M. and Yang, X. M. erythroid differentiation. Genes Dev. 24, 277-289. (2001). Isolation and characterization of EDAG-1, a novel gene related to Soza-Ried, C., Hess, I., Netuschil, N., Schorpp, M. and Boehm, T. (2010). regulation in hematopoietic system. Sheng Wu Hua Xue Yu Sheng Wu Wu Li Xue Essential role of c-myb in definitive hematopoiesis is evolutionarily conserved. Bao (Shanghai) 33, 641-646. Proc. Natl. Acad. Sci. USA 107, 17304-17308. Lu, J., Xu, W. X., Wang, S. Y., Jiang, Y., Li, C. Y., Cai, W. M. and Yang, X. M. (2002). Suzuki, M., Moriguchi, T., Ohneda, K. and Yamamoto, M. (2009). Differential [Overexpression of EDAG-1 in NIH3T3 cells leads to malignant transformation]. contribution of the Gata1 gene hematopoietic enhancer to erythroid differentiation. Sheng Wu Hua Xue Yu Sheng Wu Wu Li Xue Bao (Shanghai) 34,95-98. Mol. Cell. Biol. 29, 1163-1175. Lyons, S. E., Lawson, N. D., Lei, L., Bennett, P. E., Weinstein, B. M. and Liu, P. P. Talbot, J. C. and Amacher, S. L. (2014). A streamlined CRISPR pipeline to reliably (2002). A nonsense mutation in zebrafish gata1 causes the bloodless phenotype generate zebrafish frameshifting alleles. Zebrafish 11, 583-585. in vlad tepes. Proc. Natl. Acad. Sci. USA 99, 5454-5459. Traver, D., Paw, B. H., Poss, K. D., Penberthy, W. T., Lin, S. and Zon, L. I. (2003). Marchler-Bauer, A., Derbyshire, M. K., Gonzales, N. R., Lu, S., Chitsaz, F., Geer, Transplantation and in vivo imaging of multilineage engraftment in zebrafish L. Y., Geer, R. C., He, J., Gwadz, M., Hurwitz, D. I. et al. (2015). CDD: NCBI’s bloodless mutants. Nat. Immunol. 4, 1238-1246. conserved domain database. Nucleic Acids Res. 43, D222-D226. Volkmann, K., Rieger, S., Babaryka, A. and Köster, R. W. (2008). The zebrafish McDevitt, M. A., Fujiwara, Y., Shivdasani, R. A. and Orkin, S. H. (1997). An cerebellar rhombic lip is spatially patterned in producing granule cell populations upstream, DNase I hypersensitive region of the hematopoietic-expressed transcription factor GATA-1 gene confers developmental specificity in of different functional compartments. Dev. Biol. 313, 167-180. transgenic mice. Proc. Natl. Acad. Sci. USA 94, 7976-7981. Wakabayashi, J., Yomogida, K., Nakajima, O., Yoh, K., Takahashi, S., Engel, McGuckin, C. P., Forraz, N. and Liu, W. M. (2003). Diaminofluorene stain detects J. D., Ohneda, K. and Yamamoto, M. (2003). GATA-1 testis activation region is erythroid differentiation in immature haemopoietic cells treated with EPO, IL-3, essential for Sertoli cell-specific expression of GATA-1 gene in transgenic mouse. SCF, TGFbeta1, MIP-1alpha and IFNgamma. Eur. J. Haematol. 70, 106-114. Genes Cells 8, 619-630. Medvinsky, A., Rybtsov, S. and Taoudi, S. (2011). Embryonic origin of the adult Wells, M., Tidow, H., Rutherford, T. J., Markwick, P., Jensen, M. R., Mylonas, E., hematopoietic system: advances and questions. Development 138, 1017-1031. Svergun, D. I., Blackledge, M. and Fersht, A. R. (2008). Structure of tumor Montague, T. G., Cruz, J. M., Gagnon, J. A., Church, G. M. and Valen, E. (2014). suppressor p53 and its intrinsically disordered N-terminal transactivation domain. CHOPCHOP: a CRISPR/Cas9 and TALEN web tool for genome editing. Nucleic Proc. Natl. Acad. Sci. USA 105, 5762-5767. Acids Res. 42, W401-W407. Westerfield, M. (2000). The Zebrafish Book. A Guide for the Laboratory Use of Murayama, E., Kissa, K., Zapata, A., Mordelet, E., Briolat, V., Lin, H.-F., Handin, R. I. Zebrafish (Danio rerio), 4th edn. Eugene: Univ. of Oregon Press. and Herbomel, P. (2006). Tracing hematopoietic precursor migration to successive Wu, S. and Zhang, Y. (2007). LOMETS: a local meta-threading-server for protein hematopoietic organs during zebrafish development. Immunity 25, 963-975. structure prediction. Nucleic Acids Res. 35, 3375-3382. Nakata, T., Ishiguro, M., Aduma, N., Izumi, H. and Kuroiwa, A. (2013). Chicken Xu, J. and Zhang, Y. (2010). How significant is a protein structure similarity with TM- hemogen homolog is involved in the chicken-specific sex-determining score=0.5? Bioinformatics 26, 889-895. mechanism. Proc. Natl. Acad. Sci. USA 110, 3417-3422. Yang, L. V., Nicholson, R. H., Kaplan, J., Galy, A. and Li, L. (2001). Hemogen is a Near, T. J., Parker, S. K. and Detrich, H. W. (2006). A genomic fossil reveals key novel nuclear factor specifically expressed in mouse hematopoietic development steps in hemoglobin loss by the antarctic icefishes. Mol. Biol. Evol 23, 2008-2016. and its human homologue EDAG maps to chromosome 9q22, a region containing Nishikawa, K., Kobayashi, M., Masumi, A., Lyons, S. E., Weinstein, B. M., Liu, breakpoints of hematological neoplasms. Mech. Dev. 104, 105-111. P. P. and Yamamoto, M. (2003). Self-association of Gata1 enhances Yang, L. V., Heng, H. H., Wan, J., Southwood, C. M., Gow, A. and Li, L. (2003). transcriptional activity in vivo in zebra fish embryos. Mol. Cell. Biol. 23, 8295-8305. Alternative promoters and polyadenylation regulate tissue-specific expression of Onodera, K., Takahashi, S., Nishimura, S., Ohta, J., Motohashi, H., Yomogida, Hemogen isoforms during hematopoiesis and spermatogenesis. Dev. Dyn. 228, K., Hayashi, N., Engel, J. D. and Yamamoto, M. (1997). GATA-1 transcription is 606-616. controlled by distinct regulatory mechanisms during primitive and definitive Yang, L. V., Wan, J., Ge, Y., Fu, Z., Kim, S. Y., Fujiwara, Y., Taub, J. W., Matherly, erythropoiesis. Proc. Natl. Acad. Sci. USA 94, 4487-4492. L. H., Eliason, J. and Li, L. (2006). The GATA site-dependent hemogen promoter Osborne, C. S., Chakalova, L., Brown, K. E., Carter, D., Horton, A., Debrand, E., is transcriptionally regulated by GATA1 in hematopoietic and leukemia cells. Goyenechea, B., Mitchell, J. A., Lopes, S., Reik, W. et al. (2004). Active genes Leukemia 20, 417-425. dynamically colocalize to shared sites of ongoing transcription. Nat. Genet. 36, Yang, J., Yan, R., Roy, A., Xu, D., Poisson, J. and Zhang, Y. (2015). The I- 1065-1071. TASSER Suite: protein structure and function prediction. Nat. Methods 12, 7-8. Paffett-Lugassy, N. N. and Zon, L. I. (2005). Analysis of hematopoietic Yang, S., Ott, C. J., Rossmann, M. P., Superdock, M., Zon, L. I. and Zhou, Y. development in the zebrafish. Methods Mol. Med 105, 171-198. (2016). Chromatin immunoprecipitation and an open chromatin assay in zebrafish Piskacek, M., Havelka, M., Rezacova, M. and Knight, A. (2016). The 9aaTAD erythrocytes. Method Cell Biol. 135, 387-412. transactivation domains: from Gal4 to p53). PLoS ONE 11, e0162842. Yates, A., Akanni, W., Amode, M. R., Barrell, D., Billis, K., Carvalho-Silva, D., Postlethwait, J. H., Woods, I. G., Ngo-Hazelett, P., Yan, Y. L., Kelly, P. D., Chu, F., Cummins, C., Clapham, P., Fitzgerald, S., Gil, L. et al. (2016). Ensembl 2016. Huang, H., Hill-Force, A. and Talbot, W. S. (2000). Zebrafish comparative Nucleic Acids Res. 44, D710-D716. genomics and the origins of vertebrate . Genome Res. 10, 1890-1902. Yergeau, D. A., Cornell, C. N., Parker, S. K., Zhou, Y. and Detrich, H. W. (2005). Ransom, D. G., Haffter, P., Odenthal, J., Brownlie, A., Vogelsang, E., Kelsh, bloodthirsty, an RBCC/TRIM gene required for erythropoiesis in zebrafish. Dev. R. N., Brand, M., van Eeden, F. J., Furutani-Seiki, M., Granato, M. et al. (1996). Biol. 283, 97-112. Characterization of zebrafish mutants with defects in embryonic hematopoiesis. Zeng, Y., Xu, J., Li, D., Li, L., Wen, Z. and Qu, J. Y. (2012). Label-free in vivo flow Development 123, 311-319. cytometry in zebrafish using two-photon autofluorescence imaging. Opt. Lett. 37, Reese, M. G. (2001). Application of a time-delay neural network to promoter annotation in the Drosophila melanogaster genome. Comput. Chem. 26, 51-56. 2490-2492. Rhodes, M. M., Kopsombut, P., Bondurant, M. C., Price, J. O. and Koury, M. J. Zhang, Y. and Skolnick, J. (2005). TM-align: a protein structure alignment (2005). Bcl-x(L) prevents apoptosis of late-stage erythroblasts but does not algorithm based on the TM-score. Nucleic Acids Res. 33, 2302-2309. mediate the antiapoptotic effect of erythropoietin. Blood 106, 1857-1863. Zhang, M.-J., Ding, Y.-L., Xu, C.-W., Yang, Y., Lian, W.-X., Zhan, Y.-Q., Li, W., Xu, Rodrıguez-Maŕ ı,́ A., Yan, Y.-L., BreMiller, R. A., Wilson, C., Cañestro, C. and W.-X., Yu, M., Ge, C.-H. et al. (2012). Erythroid differentiation-associated gene Postlethwait, J. H. (2005). Characterization and expression pattern of zebrafish interacts with NPM1 (nucleophosmin/B23) and increases its protein stability, Anti-Müllerian hormone (Amh) relative to sox9a, sox9b, and cyp19a1a, during resisting cell apoptosis. FEBS J. 279, 2848-2862. gonad development. Gene Expr. Patterns 5, 655-667. Zhao, Y., Ratnayake-Lecamwasam, M., Parker, S. K., Cocca, E., Camardella, L., Rossi, A., Kontarakis, Z., Gerri, C., Nolte, H., Hölper, S., Krüger, M. and Stainier, di Prisco, G. and Detrich, H. W. (1998). The major adult α-globin gene of D. Y. R. (2015). Genetic compensation induced by deleterious mutations but not Antarctic teleosts and its remnants in the hemoglobinless icefishes: calibration of gene knockdowns. Nature 524, 230-233. the mutational clock for nuclear genes. J. Biol. Chem 273, 14745-14752. Schoenfelder, S., Sexton, T., Chakalova, L., Cope, N. F., Horton, A., Andrews, Zheng, W.-W., Dong, X.-M., Yin, R.-H., Xu, F.-F., Ning, H.-M., Zhang, M.-J., Xu, C.- S., Kurukuti, S., Mitchell, J. A., Umlauf, D., Dimitrova, D. S. et al. (2010). W., Yang, Y., Ding, Y.-L., Wang, Z.-D. et al. (2014). EDAG positively regulates Preferential associations between co-regulated genes reveal a transcriptional erythroid differentiation and modifies GATA1 acetylation through recruiting p300.

interactome in erythroid cells. Nat. Genet. 42, 53-61. Stem Cells 32, 2278-2289. Biology Open

14